sent-conv-torch icon indicating copy to clipboard operation
sent-conv-torch copied to clipboard

Published 20 hours ago • verified publisher icon harvardnlp

training error

Open Vincechen611 opened this issue 9 years ago • 6 comments
trafficstars

I got this error when I run training (main.lua). I am sure MR.hdf5 file is exist. Can someone give me some suggest.... Thanks a lot !!!

Aspire-VN7-791:~/NLP/sent-conv-torch-master$ th main.lua -data MR.hdf5 -cudnn 1 loading data... data loaded!
vocab size: 18766
vec size: 300 ==> fold 1 /home/swpc/torch/install/bin/luajit: main.lua:106: attempt to call global 'get_layer' (a nil value) stack traceback: main.lua:106: in function 'build_model' main.lua:179: in function 'train_loop' main.lua:342: in function 'main' main.lua:355: in main chunk [C]: in function 'dofile' ...swpc/torch/install/lib/luarocks/rocks/trepl/scm-1/bin/th:145: in main chunk [C]: at 0x00405e90

Vincechen611 avatar Aug 04 '16 02:08 Vincechen611

Check that you have the latest code version? It seems to work for me.

jeffreyling avatar Aug 04 '16 03:08 jeffreyling

Hi @jeffreyling Yes i have download the latest code version from github. But the issue is still exist...@@ I have install nn, hdf5, optim, nngraph, cudnn, cutorch, cunn with luarocks. Is there any package i miss? My OS is Ubuntu 16.04

Thanks for your reply!

Vincechen611 avatar Aug 04 '16 05:08 Vincechen611

Hi @jeffreyling I have solved the problem, it cause by nn version..... but i got the new problem below.....

wirobot@wirobot-Inspiron-17-7778:~/sent-conv-torch-master$ th main.lua -data MR.hdf5 -cudnn 1 loading data... data loaded!
vocab size: 18766
vec size: 300 ==> fold 1
/home/wirobot/torch/install/bin/luajit: invalid arguments: CudaTensor CudaTensor CudaTensor number expected arguments: CudaTensor | [CudaTensor] [CudaLongTensor] CudaTensor index stack traceback: [C]: at 0x7f023dd21840 [C]: in function 'max' /home/wirobot/torch/install/share/lua/5.1/nn/Max.lua:30: in function 'func' .../wirobot/torch/install/share/lua/5.1/nngraph/gmodule.lua:345: in function 'neteval' .../wirobot/torch/install/share/lua/5.1/nngraph/gmodule.lua:380: in function 'forward' ./trainer.lua:61: in function 'opfunc' ...e/wirobot/torch/install/share/lua/5.1/optim/adadelta.lua:31: in function 'optim_method' ./trainer.lua:86: in function 'train' main.lua:206: in function 'train_loop' main.lua:352: in function 'main' main.lua:365: in main chunk [C]: in function 'dofile' ...obot/torch/install/lib/luarocks/rocks/trepl/scm-1/bin/th:145: in main chunk [C]: at 0x004065c0

Vincechen611 avatar Aug 04 '16 13:08 Vincechen611

Hi, can you try updating your torch, nn, and cunn again? I noticed there were some version incompatibility issues.

jeffreyling avatar Aug 17 '16 23:08 jeffreyling

Hi, I am getting the same error. Did you guys solve this already?

adityaSomak avatar Sep 20 '16 03:09 adityaSomak

Do you have the latest torch, nn, and cunn versions?

jeffreyling avatar Sep 20 '16 19:09 jeffreyling